基于GLOBUS的网格服务实现及资源调度仿真
摘要
网格计算是分布式计算中一个重要的新领域,它的目标是通过对各种分布式的、异构的、动态的计算资源的集成,向用户提供随处可得的、灵活的、可靠的、一致的、标准的、廉价的计算能力。
搭建网格基础设施平台、实现网格服务的发布与开发以及资源管理都是目前网格计算技术的热点问题。要实现一个真实的网格服务平台需要的开销很大,而且管理起来也很复杂,因此对一般的研究单位并不适用。而局域网网格因其开销小、管理方便日渐受人青睐。另外,由于网格资源本身具有的特点,使得资源管理成为网格技术中一件非常复杂和具有挑战性的工作。资源管理的好坏、效率的高低直接关系到计算网格系统的性能,因此对资源管理的研究显得极为重要。
本文首先介绍了网格计算的基本概念、特征及其研究现状和应用领域,详细分析了五层沙漏模型、OGSA(Open Grid Service Architecture)的设计以及OGSI(Open Grid Service Infrastructure)和它的重构—WSRF(Web ServiceResource Framework)等网格体系结构,利用Globus Toolkit4.0网格计算工具包分别在windows XP和Linux操作系统上实现了单机网格平台的搭建和服务实现,然后使用这两种不同的操作系统构建了一个局域网网格平台,并在已构建的网格平台上采用java编程语言实现了MathService的发布与开发。
其次,针对使用真实网格平台来研究资源调度昂贵、耗时而且不易操作这一问题,本文提出用GridSim工具集进行仿真网格实验,并在仿真网格上实现了当前主要用于资源调度的时间最优、代价最优、时间代价均衡算法的模拟。在已有算法的基础上,本文还对均衡算法进行了改进,提出一种新的基于时间代价最优的调度算法,最后借助Eclipse平台实现了新算法的模拟。实验结果表明,改进算法在任务数比较多时具有一定的优势,使用模拟工具模拟网格环境来测试调度算法的性能方便可性。
Grid computing is an important new domain of the distributed computing. Its goal is connecting many kinds of geography distributes, the isomerism and dynamic computations resources though the high speed network to provide users with computation ability which is everywhere may result in, nimbly, reliable, consistent, standard and inexpensive.
The building of grid infrastructure platform and the realization grid service issue and the development as well as the resource management all are the present grid computing technology hot topics. The expenses to realize a real grid service platform which needs to be very big, moreover, managing it is very complex, therefore it is not suitable to the general research unit. But local area network grid receives human's favor day after day because of its small expenses and convenient management. Moreover, the characteristic which grid resource itself has causes in the resource management into grid extremely complex and challenging. The quality and the efficiency of the resource management directly relate the performance of the computing grid system, therefore it appears important to the resource management research.
Firstly, this article has introduced grid computing about basic concept, characteristic, research present situation as well as application domain. Then, the article has analysised the grid architectures such as five hour glass models, OGSA (Open Grid Service Architecture), the design of OGSI (Open Grid Service Infrastructure) and its restructure, WSRF (Web Service Resource Framework). After that, it used Globus Toolkit4.0, the grid computation tool software, separately build the grid platform in Windows XP and Linux operating system. Next, it constructed a local area network grid platform with two computers which have different operating system. Finally, in this grid computing platform, a grid application which is the classical example of grid service, MathService's development and realization have been carried on with java programming language.
Secondly, using a real grid platform for researching grid resources schedule is expensive and consumes much time, moreover, it is not easy to operate. For that, this article proposed that carrying on the simulation grid experiment with the GridSim tool collection. In this way, this article has realized three kinds of scheduling algorithm which have mainly used in the scheduling of resource, including optimise time, optimise cost as well as optimise cost and time on this simulation grid. Then the article made the improvement to the optimise cost and time algorithm and proposed a new algorithm base on time and cost. Finally it realized the new algorithm simulation with the help of the Eclipse platform. The experimental result has indicated that the improved algorithm had certain superiority when the duty number achieves certain quantity and using simulation tool to simulate grid environment and dispatching algorithm's performance is convenience and possible.
搭建网格基础设施平台、实现网格服务的发布与开发以及资源管理都是目前网格计算技术的热点问题。要实现一个真实的网格服务平台需要的开销很大,而且管理起来也很复杂,因此对一般的研究单位并不适用。而局域网网格因其开销小、管理方便日渐受人青睐。另外,由于网格资源本身具有的特点,使得资源管理成为网格技术中一件非常复杂和具有挑战性的工作。资源管理的好坏、效率的高低直接关系到计算网格系统的性能,因此对资源管理的研究显得极为重要。
本文首先介绍了网格计算的基本概念、特征及其研究现状和应用领域,详细分析了五层沙漏模型、OGSA(Open Grid Service Architecture)的设计以及OGSI(Open Grid Service Infrastructure)和它的重构—WSRF(Web ServiceResource Framework)等网格体系结构,利用Globus Toolkit4.0网格计算工具包分别在windows XP和Linux操作系统上实现了单机网格平台的搭建和服务实现,然后使用这两种不同的操作系统构建了一个局域网网格平台,并在已构建的网格平台上采用java编程语言实现了MathService的发布与开发。
其次,针对使用真实网格平台来研究资源调度昂贵、耗时而且不易操作这一问题,本文提出用GridSim工具集进行仿真网格实验,并在仿真网格上实现了当前主要用于资源调度的时间最优、代价最优、时间代价均衡算法的模拟。在已有算法的基础上,本文还对均衡算法进行了改进,提出一种新的基于时间代价最优的调度算法,最后借助Eclipse平台实现了新算法的模拟。实验结果表明,改进算法在任务数比较多时具有一定的优势,使用模拟工具模拟网格环境来测试调度算法的性能方便可性。
Grid computing is an important new domain of the distributed computing. Its goal is connecting many kinds of geography distributes, the isomerism and dynamic computations resources though the high speed network to provide users with computation ability which is everywhere may result in, nimbly, reliable, consistent, standard and inexpensive.
The building of grid infrastructure platform and the realization grid service issue and the development as well as the resource management all are the present grid computing technology hot topics. The expenses to realize a real grid service platform which needs to be very big, moreover, managing it is very complex, therefore it is not suitable to the general research unit. But local area network grid receives human's favor day after day because of its small expenses and convenient management. Moreover, the characteristic which grid resource itself has causes in the resource management into grid extremely complex and challenging. The quality and the efficiency of the resource management directly relate the performance of the computing grid system, therefore it appears important to the resource management research.
Firstly, this article has introduced grid computing about basic concept, characteristic, research present situation as well as application domain. Then, the article has analysised the grid architectures such as five hour glass models, OGSA (Open Grid Service Architecture), the design of OGSI (Open Grid Service Infrastructure) and its restructure, WSRF (Web Service Resource Framework). After that, it used Globus Toolkit4.0, the grid computation tool software, separately build the grid platform in Windows XP and Linux operating system. Next, it constructed a local area network grid platform with two computers which have different operating system. Finally, in this grid computing platform, a grid application which is the classical example of grid service, MathService's development and realization have been carried on with java programming language.
Secondly, using a real grid platform for researching grid resources schedule is expensive and consumes much time, moreover, it is not easy to operate. For that, this article proposed that carrying on the simulation grid experiment with the GridSim tool collection. In this way, this article has realized three kinds of scheduling algorithm which have mainly used in the scheduling of resource, including optimise time, optimise cost as well as optimise cost and time on this simulation grid. Then the article made the improvement to the optimise cost and time algorithm and proposed a new algorithm base on time and cost. Finally it realized the new algorithm simulation with the help of the Eclipse platform. The experimental result has indicated that the improved algorithm had certain superiority when the duty number achieves certain quantity and using simulation tool to simulate grid environment and dispatching algorithm's performance is convenience and possible.
引文
[1]都志辉,陈渝,刘鹏编著,网格计算[M].清华出版社出版,2003.05
[2]Rajkumar Buyya编,郑伟民,在威,汪东升等译,高性能集群计算:结构与系统(第一卷),电子工业出版社,272-274
[3]张锦辉.基于Agent的网格计算及其应用技术研究,硕士论文,南京航空航天大学,2004.03
[4]赵东.网格计算环境中若干关键技术的研究,硕士论文,电子科技学,2003.04
[5]徐志伟,冯百明,李伟,网格计算技术,电子工业出版社,2004.05
[6]都志辉,陈渝,刘鹏,王小鸽等,以服务为中心的网格体系结构OGSA清华大学计算机系
[7]I.Foster.网格演讲材料[EB/OL]
[8]Globus网格联盟组织[EB/OL]http://www.Globus.org http://www-fp.mcs.anl.gov/%TEfoster/talks/wwwgridsmay2002.ppt
[9]中国网格信息中转站http://www.chinagrid.net
[10]刘国恩.基于Globus工具包的网格计算系统研究,硕士论文,北京航空航天大学,2004.03
[11]Maozhenli,Mark Baker,网格计算核心技术[M].清华大学出版社.2006,17
[12]I.FosterandC.Kesselman,editors,TheGrid:BlueprintforaNew ComPutingInfrastrure.MorganKaufmann,1998.ChaPter6
[13]华宇,Grid Architecture based on WSRF,http://grid.whu.edu.cn/huayu/浅析网格体系结构一v2.ppt
[14]陈萍,余华山,王彬,网格计算环境Globus介绍,计算机应用研究,2003.08,96-98
[15]罗荣.面向服务的企业网格计算基础架构设计:硕士论文,武汉大学,2004
[16]OGSA规范,http://www.gridforum.org/ogsi-wg/drafts/Gs Spec-draft03
[17]Ian Foster,Carl Kesselman.The Grid 2:Blueprint for a New Computing Infrastructure,[M].Beijing:Electronics Industry,2004.
[18]金海,ChinaGrid Overview,http://www.chinagrid.net/grid/paperppt/bigfile /Conference/ISIT/jh-CbinaGrid-Overview.ppt
[19]姜永发,闾国年,网格计算与Grid GIS体系结构与关键技术探讨.测绘科学第4 期,2005年8月
[20]冯建红,谢汶,基于Globus Toolkit4实现网格服务,科技咨询,2006(6):87-882002-07-17.pdf
[21]Web Service工作组,http://www.w3.org/2002/ws/
[22]D.C ruspipi,K.kolarov,okhatib,the haptic display of complex graphical Environments,[j].proc of ACMSIGGRAPH,1997
[23]桂小林.网格技术导轮.北京邮电大学出版社,2005
[24]C.Lee,C.Kesselman,J.Stepanek,R.Lindell,I.Foster.The Quality of Service
[25]David Fernández-Baca.Allocating modules to processors in a distributed system.IEEE Transactions on Software Engineering,November,1989,15(11):1427-1436.
[26]BRAUN TD,SIEGEL HJ,BECK N,A Comparison of Eleven Static Heuristics for Mapping a Class of Independent Tasks onto Heterogeneous Distributed Computing Systems,[J].Journal of Parallel and Distributed Computing,2001,61(1):810-837.
[27]Ian Foster,Carl Kesselman,JeffreyMNick,etal.The physiology of the grid [C].International[J].Supercomputer Applications,2003
[28]R.Buyya,D.Abramson,and J.Giddy,An Economy Driven Resource Management Architecture for Global Computational Power Grids,Proceedings of the 2000Internationl Conference on Parallel
[29]BHARADWAJ V,GHOSED,ROBERTAZZITG.Divisible Load Theory:A New Paradigm for Load Scheduling in Distributed Systems,[J].Cluster Comput.,2003,6(1):7-17
[30]吴淞,基于网格仿真平台GRIDSIM的任务调度算法:硕士论文,四川大学,2006
[31]Aida K,Takefusa A,Nakada Hetal.《Performance Evaluation Model for Scheduling in a Global Computing System》[J].The International Journal of High Performance Computing Applications,2000,14(3):17-19
[32]R.Buyya,M.Murshed.GridSim:A Toolkit for the Moddeling and Simulation of Distributed Resource Management and Scheduling for Grid Computing.The Journal of Concurrency and Computation:Practice and Experience,Vol.14,Issue 13-15. Wiley Press,2002
[33]刘祥瑞,朱建勇,樊孝忠.基于GridSim的网格调度模拟
[34]秦金磊,网格计算环境下资源管理的研究,硕士论文,华北电力大学,2006
[35]郑然,李胜利,金海.网格资源管理与调度模型的研究.华中科技大学学报(自然科学版),2001,29(12):P87-89
[36]华中科技大学CGCL实验室http://grid.hust.edu.cn/index.html
[37]田鸣华.网格环境下的资源监控系统的研究与实现,硕士论文.国防科技大学,2004
[38]M.Ponamgi,D.Manocha,M.Lin.Incremental algorithms for collision detection between polygonal models,[J].IEEE Transactions on Visualization and Computer Graphics,1997
[39]Song HJ,Liu X,Jakobsen Detal.The MicroGrid:A Scientific Tool for Modeling Computational Grids[C].Proceedings of the 2000 Conference on Supercomputing,2000:4-10
[40]Abramson D,Giddy J,Kotler L.High.Performance Parametric Modeling with Nimrod/G:Killer Application for the Global Grid.Proceedings of the 14~(th)International Parallel&Distributed Processing Symposium.Cancun,Mexico,2000:P520-528
[41]William H.Bell,David G.Cameron,Luigi Capozza,A.Paul Millar,Kurt Stockinger,Floriano Zini.OptorSim-A Grid Simulator for Studying Dynamic Data Replication Strategies.International Journal of High Performance Computing Applications,2003
[42]刘一萌,基于GridSim网格资源管理调度算法研究,硕士论文,四川大学,2006
[43]Houda Lamehamedi,Zujun Shentu,Boleslaw Szymanski,Ewa Deelman.Simulation of Dynamic Data Replication Strategies in Data Grids.Proceeding of 12~(th)Heterogeneous Computing Workshop,2003.
[2]Rajkumar Buyya编,郑伟民,在威,汪东升等译,高性能集群计算:结构与系统(第一卷),电子工业出版社,272-274
[3]张锦辉.基于Agent的网格计算及其应用技术研究,硕士论文,南京航空航天大学,2004.03
[4]赵东.网格计算环境中若干关键技术的研究,硕士论文,电子科技学,2003.04
[5]徐志伟,冯百明,李伟,网格计算技术,电子工业出版社,2004.05
[6]都志辉,陈渝,刘鹏,王小鸽等,以服务为中心的网格体系结构OGSA清华大学计算机系
[7]I.Foster.网格演讲材料[EB/OL]
[8]Globus网格联盟组织[EB/OL]http://www.Globus.org http://www-fp.mcs.anl.gov/%TEfoster/talks/wwwgridsmay2002.ppt
[9]中国网格信息中转站http://www.chinagrid.net
[10]刘国恩.基于Globus工具包的网格计算系统研究,硕士论文,北京航空航天大学,2004.03
[11]Maozhenli,Mark Baker,网格计算核心技术[M].清华大学出版社.2006,17
[12]I.FosterandC.Kesselman,editors,TheGrid:BlueprintforaNew ComPutingInfrastrure.MorganKaufmann,1998.ChaPter6
[13]华宇,Grid Architecture based on WSRF,http://grid.whu.edu.cn/huayu/浅析网格体系结构一v2.ppt
[14]陈萍,余华山,王彬,网格计算环境Globus介绍,计算机应用研究,2003.08,96-98
[15]罗荣.面向服务的企业网格计算基础架构设计:硕士论文,武汉大学,2004
[16]OGSA规范,http://www.gridforum.org/ogsi-wg/drafts/Gs Spec-draft03
[17]Ian Foster,Carl Kesselman.The Grid 2:Blueprint for a New Computing Infrastructure,[M].Beijing:Electronics Industry,2004.
[18]金海,ChinaGrid Overview,http://www.chinagrid.net/grid/paperppt/bigfile /Conference/ISIT/jh-CbinaGrid-Overview.ppt
[19]姜永发,闾国年,网格计算与Grid GIS体系结构与关键技术探讨.测绘科学第4 期,2005年8月
[20]冯建红,谢汶,基于Globus Toolkit4实现网格服务,科技咨询,2006(6):87-882002-07-17.pdf
[21]Web Service工作组,http://www.w3.org/2002/ws/
[22]D.C ruspipi,K.kolarov,okhatib,the haptic display of complex graphical Environments,[j].proc of ACMSIGGRAPH,1997
[23]桂小林.网格技术导轮.北京邮电大学出版社,2005
[24]C.Lee,C.Kesselman,J.Stepanek,R.Lindell,I.Foster.The Quality of Service
[25]David Fernández-Baca.Allocating modules to processors in a distributed system.IEEE Transactions on Software Engineering,November,1989,15(11):1427-1436.
[26]BRAUN TD,SIEGEL HJ,BECK N,A Comparison of Eleven Static Heuristics for Mapping a Class of Independent Tasks onto Heterogeneous Distributed Computing Systems,[J].Journal of Parallel and Distributed Computing,2001,61(1):810-837.
[27]Ian Foster,Carl Kesselman,JeffreyMNick,etal.The physiology of the grid [C].International[J].Supercomputer Applications,2003
[28]R.Buyya,D.Abramson,and J.Giddy,An Economy Driven Resource Management Architecture for Global Computational Power Grids,Proceedings of the 2000Internationl Conference on Parallel
[29]BHARADWAJ V,GHOSED,ROBERTAZZITG.Divisible Load Theory:A New Paradigm for Load Scheduling in Distributed Systems,[J].Cluster Comput.,2003,6(1):7-17
[30]吴淞,基于网格仿真平台GRIDSIM的任务调度算法:硕士论文,四川大学,2006
[31]Aida K,Takefusa A,Nakada Hetal.《Performance Evaluation Model for Scheduling in a Global Computing System》[J].The International Journal of High Performance Computing Applications,2000,14(3):17-19
[32]R.Buyya,M.Murshed.GridSim:A Toolkit for the Moddeling and Simulation of Distributed Resource Management and Scheduling for Grid Computing.The Journal of Concurrency and Computation:Practice and Experience,Vol.14,Issue 13-15. Wiley Press,2002
[33]刘祥瑞,朱建勇,樊孝忠.基于GridSim的网格调度模拟
[34]秦金磊,网格计算环境下资源管理的研究,硕士论文,华北电力大学,2006
[35]郑然,李胜利,金海.网格资源管理与调度模型的研究.华中科技大学学报(自然科学版),2001,29(12):P87-89
[36]华中科技大学CGCL实验室http://grid.hust.edu.cn/index.html
[37]田鸣华.网格环境下的资源监控系统的研究与实现,硕士论文.国防科技大学,2004
[38]M.Ponamgi,D.Manocha,M.Lin.Incremental algorithms for collision detection between polygonal models,[J].IEEE Transactions on Visualization and Computer Graphics,1997
[39]Song HJ,Liu X,Jakobsen Detal.The MicroGrid:A Scientific Tool for Modeling Computational Grids[C].Proceedings of the 2000 Conference on Supercomputing,2000:4-10
[40]Abramson D,Giddy J,Kotler L.High.Performance Parametric Modeling with Nimrod/G:Killer Application for the Global Grid.Proceedings of the 14~(th)International Parallel&Distributed Processing Symposium.Cancun,Mexico,2000:P520-528
[41]William H.Bell,David G.Cameron,Luigi Capozza,A.Paul Millar,Kurt Stockinger,Floriano Zini.OptorSim-A Grid Simulator for Studying Dynamic Data Replication Strategies.International Journal of High Performance Computing Applications,2003
[42]刘一萌,基于GridSim网格资源管理调度算法研究,硕士论文,四川大学,2006
[43]Houda Lamehamedi,Zujun Shentu,Boleslaw Szymanski,Ewa Deelman.Simulation of Dynamic Data Replication Strategies in Data Grids.Proceeding of 12~(th)Heterogeneous Computing Workshop,2003.